Optimized ANN-Based Methodology for Fault Detection and Localization in Power Transmission Networks

Abstract

Transmission lines are integral to transporting electrical power from generation sites to consumers. Transmission lines are subject to various faults that disrupt service and threaten system integrity. Fault analysis (identification, classification, and localization) is essential to minimize downtime and operational costs. Improved fault control raises grid dependability, decreases outages, and optimizes operations, promoting renewable integration and cost savings. It enhances safety, power quality, and resilience while facilitating innovative grid modernization and scalability for future demands. Such developments strengthen consumer trust and lead to more sustainable, efficient, and resilient power systems. This research employs Artificial Neural Networks (ANN) to enhance fault detection on high-voltage transmission lines. Simulations were conducted on a 132 kV, 50 Hz, 100 km transmission line model using MATLAB/Simulink, generating data from various fault scenarios. The ANNs, trained with these datasets, effectively and accurately analyzed the faults. The most effective neural network architecture was identified, assuring dependable operation in various fault scenarios and showcasing a strong strategy to enhance power transmission efficiency. Configuration 2 achieved the best fault identification accuracy of 97.99%, demonstrating the system's low error rate in accurately detecting the flaw. Fault classification with Configuration 1 attained a 95.65% accuracy rate. This indicates that the system can effectively categorize various fault types. The fault location was at an accuracy of 94.51% using Configuration 1.